PM36J6B10-01E_25 Yokogawa Electric Corporation 2-9-32 Nakacho, Musashino-shi Tokyo, 180-8750 JAPAN Public Relations Dept., Yokogawa Electric Corporation.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
PM36J6B10-01E_25
Yokogawa Electric Corporation 2-9-32 Nakacho, Musashino-shiTokyo, 180-8750JAPAN
Upper solution of process operation done by operatorsTo assist daily work of production people to realize safer and more cost-effective operation. For example,– In desk work
• Support to prepare an operation report• Support to turn up operation related problems• Support to standardize operation know-how
– In normal operation• Suppress unnecessary DCS alarms to notify only important DCS alarms• Notify an early sign of abnormality faster than it is detected by DCS
alarms• Prevent miss-operation
– In transition operation (e.g. startup/load change/reactor switchover)
• Give an adequate instruction to the operator in a sequential order• Prevent miss-step/miss-procedure
Process
CONTROL domain OPERATION domain
MES, ERP
Process Control (DCS)
Advanced Process Control Advanced Operation Assistance
APC (Advanced Process Control)– Position: Additional function of DCS– Purpose: Improvement of controllability– Configuration: Runs on additional PC– Main user: Engineering people– Approach: Based on mathematical process model– Benefit: More cost-effective control
AOA (Advanced Operation Assistance)– Position: Additional function of DCS– Purpose: Improvement of operation work– Configuration: Runs on operator’s console or additional
PC– Main user: Production people (operator/process
engineer)– Approach: Based on knowledge and experiences of
skillful operator/process engineer– Benefit: Safer and more cost-effective operation
• System alarm• Process alarm• Annunciator message• Operation guidance, etc.
– Operators actions (-)• Tag data entry• Tag mode change, etc.
No. Pattern Type EBT Pattern Suspected Problem Countermeasure
Long term analysis about seasonal change
Long term analysis about erosion/corrosion
Unnecessary alarms/messages Retuning of alarm set valuesIntegration of redundant alarms/messagesMasking of low priolity alarms/messages
Low automation rate Automation using DCSManual operation according to know-how Automation using ExapilotComplex operation sequence Simplify operation sequence
Lack of support function Navigation using Exapilot
Insufficient Operator capability Operator training
Human error Error detection using Exapilot
Unstable process Introduction of advanced process control
(1)Select “Sub Type” tub(2)Select “Tag Alarm”(3)Press “Scope in filter” button(4)Select “Tool\Export Result” to save a result of analysis as CSV file(5)Edit a report file using MS-Excel function
(1)Select “Detail” tub(2)Select “MAN”(3)Press “Scope in filter” button(4)Select “Tool\Export Result” to save a result of analysis as CSV file(5)Edit a report file using MS-Excel function
Additional DCS function for– Knowledge-based process control– Knowledge-based process monitoring
Application can be programmed as flow chart or logic chart– Specialist programming skill is not needed– High usability– High understandability– High maintainability
Existing application in DCS is no need to modify
Various usages– Automation of transient operation– Alarm management– Operator training
<Condition icon>Check P101.ALRM = NR(NR means Normal)
<Output to DCS icon>P101.CSV to 0 (Stop)
<Alarm message icon>P101 start error
<Pause icon>Pause this sequence
<Alarm message icon>Preparation error
<Output to DCS icon>FIC100.SV to 20t/h
<Timer icon>Wait 10 minutes
YES YES
NONO
NOYES
Typical paper SOP
Know-how
Know-how
(1) Check base tank level LI100.PV >= 50%(2) Start pump P-101(3) Check answer back flag(4) open field hand valve HV100
Typical paper SOP (too much simplified)
If tank level is lower than 50%, announce it and open inlet valve of base tank (FIC100.SV = 20t/h), and wait for 10 minutes. If answerback is not ‘2’ and alarm status (P101.ALRM) is not “NR”, stop pump (P101.CSV = 0), announce it, and pause operation till field operator solves the problem.
Operation know-how of veteran operator/engineer
An automation program can be easily created as flow chart diagram including a veteran operator’s/engineer’s know-how.
1. Icons’ color means the condition. Green:Now executing Gray:Finished White:Unexecuted Yellow:Paused
1 2 3 4
1 2 3 4 5 6 7 8
Flow chart diagram can be executed automaticallyInteractive operation between Exapilot and operators can be doneVarious kinds of messages makes operators right judgmentOperational status can be confirmed visuallyManual interposition by operators can be done according to need
2. Four kinds of messages display. 1)Confirm 2)Guide 3)Alarm 4)System
3. Manual operation is available. 1)Start all 2)Start icon 3)Stop icon 4)Pause (all/icon) 5)Initialize icon 6)Break icon 7)Skip icon 8)Stop all
90% of manual operations were automatedAll monitoring works were automated2000 unnecessary alarms were suppressed– Operators can do another work during crude switchover
operation– Miss-step and miss-procedure were prevented from
• Mixture of law material (Crude gas oil)• Distillation (Extraction of impurities)• Hydro treating• Extraction (Dissolution of BTX)• Recovery• Distillation (Constituent separation of BTX)
System configuration– CENTUM-XL with ABC– Exapilot on HIS
Application example– Automation of reactor depressurisation
• Extract liquid/gas to downstream in order to reduce inlet pressure when reactor is stopped
• In case of junior operator, it takes long time to depress inlet pressure for fear of process fluctuation
• In case of skilful operator, he can immediately depress inlet pressure by monitoring pressure and flow of downstream so that process load of downstream becomes upper limitation constantly
• Operation time was reduced from 5 hours to 2.5 hours– Automation of reactor pressurisation
• Operation time was reduced 30 minutes– Automation of feed control during start up operation
• Feed flow of other reactors is fluctuated according to the fluctuation of header pressure while operator ramps up load of feed compressor to start up purged reactor, and it causes mental distress to the operators due to the possibility of abnormal reaction
• Full automation of board operation was achieved• 290 manipulations were reduced in each purge operation• 410 hours/year were reduced including monitoring work
– Increase of productivity• 3 hours shortened in each purge operation
(150hours/year)• Productivity was increased 200 ton/year
– Prevention of miss-operation• Check list was eliminated
– Reduction of mental distress– Succession of operation know-how
Benefits– Reduce work load of unusual state operation– Standardise plant start up operation– 200 field activities are confirmed by Exapilot– 500 DCS parameters are monitored by Exapilot– 420 DCS parameters are automatically adjusted– Prevent miss-step or miss-procedure– Ensure timing of Plant Start Up operation
(1) Masking of unnecessary DCS alarm(2) Dynamic alarm setting(3) Addition of pre-alert(4) Replacement to advanced alarm(5) Prevention of miss-operation
For safer and more cost-effective operation of industrial systems– To help the operator to correct potentially dangerous
situations before the ESD is forced to intervene– To recognize and act to avoid hazardous situations– To identify deviations from desired operating conditions
that could lead to financial loss– To better understand complex process conditions
Unnecessary alarms notified despite safe or efficient operationNecessary alarms not notified despite unsafe or inefficient operation(it’s not general because alarm must be set narrowly for safety reasons)
(1) IF New_PH = Mean - 3 x Dev >= Original_PL + 0.1 x (SH - SL) & |New_PH - Mean| >= |Original_PH - Mean|
THEN New_PH = Mean - 3 x Dev
(2) IF New_PH = Mean - 3 x Dev <= Original_PL + 0.1 x (SH - SL) & |New_PH - Mean| >= |Original_PH - Mean|
THEN New_PH = Mean - A x Dev
SL
SH
Original_PH
Original_PL
Start monitoring
1 2 3
New_PH
Number of occurrence
HI HI HI
Time
PV
Start
masking
Purpose
This logic can mask an unnecessary HI/LO alarm which is slowly oscillating near PH/PL.
Logic
This logic consists of “monitoring” logic, “masking” logic and “recovery” logic. Monitoring logic is started when first HI/LO alarm is occurred, and then monitors the number of occurrence while ordered monitoring period. In case the number reaches an ordered times, masking logic is run.
Masking logic
PH/PL is automatically changed so that new PH/PL is not announced. The reason of suppression is displayed as a countermeasure.
Reason of occurrence: Alarm threshold is unmatched
This logic masks an unnecessary annunciator which announced repeatedly.
Masking logic
In case annunciator is announced ordered times in monitoring period, AOF is set to target annunciator (%AN). Then target tag is registered as habitual.
ON
%ANN
OFF
AON
Monitoringperiod
TimeMonitoringperiod
1 2 3
Occurrencenumber
Start
masking
Recovery logic
In case annunciator is not announced for monitoring period, AON is set to target annunciator (%AN).
Masking logic (After learning)
In case registered tag announced annunciator again, AOF is set without waiting.
Reason of occurrence: DCS sequence is not prepared
This logic detects true IOP/IOP- alarms which announced due to range over.
Detection Logic
In case IOP/IOP- is happened after HI/HH or LO/LL alarm is notified, and is recovered after ordered period, this logic shows the reason of IOP/IOP- alarm (range over). Also target tag is registered as habitual.
Exclude logic
In case registered tag announces IOP/IOP- alarm again, target tag is excluded from watch list.
HHNR HH
HH
PH
HI
IOP
Start
monitoringMonitoring periodfor IOP detection
Reason of occurrence: Range over (engineering problem)
This logic detects true IOP/IOP- alarm which notified due to the failure of field transmitter.
Detection logic
In case IOP/IOP- is happened even though HI/HH or LO/LL alarm is not notified, or IOP/IOP- is happened immediately (within 5 seconds) after HI/HH or LO/LL alarm is notified, this logic shows the reason of IOP/IOP- alarm (failure of field transmitter). Also target tag is registered as habitual.
Masking logic
In case registered tag announces IOP/IOP- alarm again, AOF is set continuously.
NRNR NR
HH
PH
IOP
JudgmentStart
monitoringStart
monitoring
SH
Monitoringperiod
Monitoring period
HI
AON
Reason of occurrence: Failure of field transmitter
This logic announces the reason of true IOP/IOP- (breaking of wire) alarms.
Detection Logic
In case IOP/IOP- is notified even though HI/HH or LO/LL alarms are not notified, and IOP/IOP- is not recovered for ordered period, this logic shows the reason of alarm notification.
– Quality monitorDetect an early sign of abnormality faster than it’s detected by DCS alarmsNotify the reason of alarm and adequate countermeasure to operatorsSend e-mail to relevant peopleExecute countermeasure automatically if necessary
Monitoring Failure of Field Instrumentation– Detect Invariant Indicated Value (A01)– Detect Continuously Increasing Indicated Value (A02)– Detect Continuously Decreasing Indicated Value (A03)– Detect Deviation Alarm between 2-instruments (A04)– Detect Deviation Alarm between 3-instruments (A05)
Monitoring Process Abnormalities– Detect Hunting S Type (Standard deviation type) (B01)– Detect Hunting DS Type (Deviation of standard deviation) (B02)– Detect Deviation Alarm between Theoretical Value and Raw Value (B03)
Monitoring Failure/Overload of Field Equipments– Monitoring Control Valve (EXA01)– Monitoring Pressure Gauge (EXA02)– Monitoring Reciprocating Compressor Abnormalities (EXA03)– Monitoring Clogging in Process Units (EXA04)
Monitoring Process Abnormalities– Monitoring Reactor Temperature Distribution (EXB01)– Monitoring Heat Balance Abnormality in the Furnace (EXB02)
This logic detects the abnormality of cylinder valve in reciprocating compressor by comparing the difference between estimated outlet temperature and measured outlet temperature.In case cylinder valve is broken or quantity of cooling water is insufficient, the outlet temperature will be risen due to back flow inside of cylinder.
Example4: Pressure indicator diagnosis This logic detects the abnormality of pressure indicator by checking the difference of three pressures.
Furnace
FI103.PV
FI101.PV
FI102.PV
TI100.PV
TI200.PVLowMaterial A
LowMaterial B
FuelGas
XI103.PV
Example5: Furnace incomplete combustion diagnosis
This logic detects early sign of incomplete combustion in furnace unit before DCS HI alarm detects it, by monitoring the difference between “Fuel gas calorie” calculated using fuel gas flow & gravity and “Furnace duty” calculated using flow & temperature of furnace unit.
Incorrect action must be detected ASAP to prevent serious troubleExapilot Advanced Alarm can detect various kinds of miss-operation by monitoring operator’s behavior
Example: Miss of tag mode change
IF FIC100.MODE is changed to “MAN” 3 minutes lapsed FIC100.MODE is still in “MAN”THENExapilot changes tag mode to “AUT”
Use Exapilot and CS 3000 test function (WDA)No need to modify existing DCS applicationLow cost systemSimple process model can be made in Exapilot– Dead time– Time lag– Material balance, etc.
General training using HIS windowsTraining for transition operation using Exapilot window